Article pubs.acs.org/est
Isotopic Analysis of Oxidative Pollutant Degradation Pathways Exhibiting Large H Isotope Fractionation Reto S. Wijker,†,‡,∥ Pawel Adamczyk,§,∥ Jakov Bolotin,† Piotr Paneth,*,§ and Thomas B. Hofstetter*,†,‡ †
Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-8600 Dübendorf, Switzerland Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zürich, CH-8092 Zürich, Switzerland § Switzerland Institute of Applied Radiation Chemistry, Technical University of Lodz, Zeromskiego 116, 90-924 Lodz, Poland ‡
S Supporting Information *
ABSTRACT: Oxidation of aromatic rings and its alkyl substituents are often competing initial steps of organic pollutant transformation. The use of compound-specific isotope analysis (CSIA) to distinguish between these two pathways quantitatively, however, can be hampered by large H isotope fractionation that precludes calculation of apparent 2H-kinetic isotope effects (KIE) as well as the process identification in multi-element isotope fractionation analysis. Here, we investigated the C and H isotope fractionation associated with the transformation of toluene, nitrobenzene, and four substituted nitrotoluenes by permanganate, MnO4−, to propose a refined evaluation procedure for the quantitative distinction of CH3group oxidation and dioxygenation. On the basis of batch experiments, an isotopomer-specific kinetic model, and density functional theory (DFT) calculations, we successfully derived the large apparent 2H-KIE of 4.033 ± 0.20 for the CH3-group oxidation of toluene from H isotope fractionation exceeding >1300‰ as well as the corresponding 13C-KIE (1.0324 ± 0.0011). Experiment and theory also agreed well for the dioxygenation of nitrobenzene, which was associated with 2H- and 13C-KIEs of 0.9410 ± 0.0030 (0.9228 obtained by DFT) and 1.0289 ± 0.0003 (1.025). Consistent branching ratios for the competing CH3-group oxidation and dioxygenation of nitrotoluenes by MnO4− were obtained from the combined modeling of concentration as well as C and H isotope signature trends. Our approach offers improved estimates for the identification of contaminant microbial and abiotic oxidation pathways by CSIA.
■
and/or aromatic ring dioxygenation, respectively.10−16 The diagnostic value of such multi-element isotope fractionation patterns is based on the fact that C and H isotope fractionation is caused by kinetic isotope effects (KIEs), which are indicative of each oxidation pathway.5,6,8,17,18 The impact of 13C- and 2HKIEs on the observable isotope fractionation, however, is diluted by nonreactive atoms in a molecule. Dilution of isotope fractionation is taken into account in bulk compound C and H isotope enrichment factors (ϵC and ϵH, respectively), which are used to quantify a contaminant’s isotope fractionation behavior.6 Correlations of changing H and C isotope signatures, δ2H vs δ13C, can be used for the identification and quantification of competing reaction pathways because the correlation slopes, Δδ2H vs Δδ13C (eq 1), follow approximately the ratio of ϵC and ϵH values.19,20
INTRODUCTION Oxidations of methyl groups and aromatic rings are frequent initial steps in the enzymatic and abiotic degradation of persistent aromatic contaminants in the environment such as nitroaromatic explosives and fuel components (BTEX).1,2 Because both processes can occur concurrently and lead to products that are susceptible for further transformation, they are difficult to track, thus making a quantitative assessment of such transformations very challenging.3 This problem was successfully alleviated by the application of compound-specific isotope analysis (CSIA), with which changes of isotope ratios in the residual contaminant have been used for quantifying the extent of transformation even if multiple chemical and/or biological reactions took place.4−9 Because oxidation of methyl groups and aromatic rings involve the cleavage of C−H bonds or CC and C−H hybridization changes, the combined analysis of changes in C and H isotope ratios is a robust measure to assess (bio)degradation without explicit need for monitoring reactant and product concentration time series. Indeed, laboratory and field studies with various substituted benzenes have revealed characteristic C and H isotope fractionation trends for contaminant degradation initiated via methyl group oxidation © 2013 American Chemical Society
Received: Revised: Accepted: Published: 13459
August 12, 2013 October 25, 2013 October 31, 2013 October 31, 2013 dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
Scheme 1. Initial Steps of MnO4− Catalyzed Oxidation Pathways of Toluene and Nitrotoluenes, Including Their C and H Atom Labels Used Throughout the Articlea
a
Water is involved in the stabilization of the CH3-group oxidation transition state.21
ϵ Δδ 2 H ≈ H ϵC Δδ13C
fractionation patterns of aliphatic and aromatic C oxygenation of six model contaminants by permanganate (MnO4−). Based on the comparison of (experimental) 2H- and 13C-AKIE values with the corresponding (computed) intrinsic KIEs, we propose a new data evaluation procedure for (i) the derivation of large 2 H-AKIEs from H isotope fractionation data as well as (ii) the interpretation of competing CH3-group oxidation and dioxygenation reactions in typical two-dimensional C vs H isotope fractionation plots for compounds that differ in the number of C and H atoms. MnO4− was chosen as model oxidant owing to its use for in situ chemical oxidation of organic contaminants31 as well as previous experimental and theoretical reports on large KIEs associated with aliphatic and olefinic C−H bond oxygenations19,21,29,32−34 (Scheme 1). The set of probe compounds included typical soil and water contaminants that react with MnO4− exclusively via CH3-group oxidation (toluene21) and aromatic ring dioxygenation (nitrobenzene), respectively, as well as four nitro- and dinitrotoluenes (2- and 4-nitrotoluene, 2,4- and 2,6-dinitrotoluene) for which both oxygenation pathways occur.
(1)
2
However, H-KIEs associated with the cleavage of aliphatic C−H bonds can be substantial (i.e., 7−5019,21−25) and eq 1 no longer holds because δ2H vs δ13C correlations become nonlinear as pointed out earlier by Elsner et al.19 This phenomenon compromises the identification of reaction pathways by CSIA in dual isotope plots. Moreover, the large H isotope fractionation arising from such reactions cannot be quantified as apparent kinetic isotope effects (AKIE). The typically used AKIE-approximation, eq 2, leads to numerically erroneous results including meaningless negative AKIE values once bulk H isotope enrichment factors, ϵH, approach the negative inverse of the number of H atoms, −1/n, in a compound.26 AKIE E ≈
1 1 + n·ϵE
(2)
where subscript E stands for isotopic elements H and C, respectively. Failure to compare intrinsic 2H-KIEs with experimental 2H-AKIEs derived from H isotope fractionation as in eq 2 impedes a mechanistic interpretation of the observed contaminant isotope fractionation27 that would be necessary for a correct application of CSIA. Another challenge regarding the interpretation of AKIEs obtained with eq 2 arises for dioxygenation pathways. Such concerted reactions exhibit two, often distinctly different primary 2 H- and 13C-KIEs due to asynchronous dioxygen addition.15,28,29 Because eq 2 returns the weighted average of two position-specific KIEs,30 a comparison of 13C-AKIEs for dioxygenation warrants theoretical support such as density functional theory-based KIE calculations for unambiguous interpretation. The goal of this study was to contribute to a precise interpretation of C and H isotope fractionation data gained from CSIA for oxidation reactions of substituted aromatic contaminants. To this end, we performed a combined experimental and computational study in which we evaluated the 2H- and 13C-kinetic isotope effects and C and H isotope
■
EXPERIMENTAL METHODS A complete list of all chemicals used and their purity can be found in the Supporting Information (SI). Batch Experiments. Oxidation experiments with toluene (1.8 mM initial concentration), 2-NT (1.9 mM), 4-NT (1.3 mM), and 2,4-DNT (0.5 mM) were carried out in batch reactors containing 50 mM KMnO4. Because of slow turnover, oxidations of nitrobenzene (1.9 mM) and 2,6-DNT (0.4 mM) were performed in 300 mM and 87.5 mM KMnO4, respectively. All reactions took place at room temperature in serum bottles closed with Viton stoppers and buffered at pH 7.2 with 25 mM potassium phosphate. Blanks were treated identically except for the addition of KMnO4 to rule out losses due to phase transfer processes. Oxidations were initiated through the addition of KMnO4 stock solution. At predefined time points, aqueous samples were withdrawn with a gastight glass syringe, and the reaction was stopped by the addition of stoichiometric excess of NaS2O3. The samples were subsequently filtered through a 0.22 13460
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
⎛ δ13C + 1 ⎞ ⎛ δ 2H + 1 ⎞ ϵ ⎟ ⎟ = H ·ln ⎜ 13 ln ⎜ 2 ϵC ⎝ δ C0 + 1 ⎠ ⎝ δ H0 + 1 ⎠
μm hydrophilic PTFE filter and stored at 4 °C until chemical and isotopic analysis. Chemical and Isotopic Analyses. Concentrations of nitrobenzene as well as mono- and dinitrotoluenes were quantified by reverse phase HPLC (Dionex UltiMade 3000) using a Supelcosil LC-18 column (25 cm × 4.6 mm, 5 μm, Supelco) and UV−vis detection at wavelengths corresponding to the absorption maxima of the compounds. The eluent mixture consisted of 65% methanol and 35% H2O at a flow rate of 1 mL min−1. Toluene was extracted from buffered aqueous solution in hexane and measured by GC/MS using methods described recently.10 Stable C, H, and N isotope signatures of toluene and nitrotoluenes were measured using a gas chromatograph coupled to an isotope ratio mass spectrometer via a combustion interface as reported previously.35−37 Briefly, extraction and enrichment from aqueous samples containing nitrotoluenes or nitrobenzene were performed by solid-phase microextraction (SPME, 85 μm polyacrylate coating, Supelco) at 40 °C for 45 min. Toluene was extracted from a purge and trap system (Teledyne Tekmar) according to the method reported in ref 38. Samples were diluted to concentrations yielding similar peak amplitudes (between 1 and 5 V) to minimize analytical uncertainties due to instrument nonlinearity. C, N, and H isotope signatures (δ13C, δ15N, and δ2H), are reported as arithmetic mean of triplicate measurements (±σ) in per mil (‰) relative to Vienna PeeDee Belemnite (δ13CVPDB), air (δ15Nair), and Vienna standard mean ocean water (δ2HVSMOW), respectively. All compounds were available as calibrated inhouse standards, whose δ 13 C and δ 15 N values were characterized independently by elemental analyzer IRMS and used in standard bracketing procedures to ensure accuracy of isotopic measurements. Hydrogen isotope ratios were calibrated using different alkanes and dimethylaniline as reference compounds39 in the δ2H-range between −50‰ to 500‰. Evaluation of Experimental Data. For the comparison of data evaluation procedures, typical bulk isotope enrichment factors, ϵE, were derived from linear regression of δ2H or δ13C values versus the remaining fraction of the reactant as shown in eq 3. ⎛ δhE + 1 ⎞ ⎛c⎞ ⎟⎟ = ϵE ·ln ⎜ ⎟ ln ⎜⎜ h ⎝ c0 ⎠ ⎝ δ E0 + 1 ⎠
ΛH/C =
ϵH ϵC
(4)
(5)
2 H- and 13C-AKIE values pertinent to CH3-group oxidation and dioxygenation were calculated from experiments with toluene and nitrobenzene as substrates, respectively. To this end, we implemented a numerical model in Aquasim41 consisting of a set of differential equations for all C and H isotopologues and isotopomers containing light and one heavy C or H isotope, respectively (eq 6).
d[ci] = −∑ γjE·kjE·[ci] dt j
AKIEEj
(6)
kjl ,E
=
kjh ,E
(7)
where [ci] is the concentration of isotopomer i (see compilation of isotopomers in section S3 of the Supporting Information), γEj is the probability of reaction j (i.e., CH3-group oxidation or dioxygenation) at a reactive position of the substrate, and kEj is the reaction’s second-order rate constant. γEj is used to account explicitly for intramolecular competition of randomly distributed heavy isotopes (Tables S4−S7, Supporting Information). Superscript E stands for the C and H isotopes, respectively. 2H- and 13C-AKIE values follow from the ratio of reaction rate constants of light (l) and heavy (h) C and H isotopomers (eq 7). The numerical model was fit to the evolution of measured isotope signatures, δhE(t), and substrate concentrations, ctot(t), over the time course of the reactions. These two variables were obtained numerically from summing up n isotopomers as well as C and H atoms (xh,E and xl,E i i ) as shown in eqs 8 and 9. − MnO4 -catalyzed oxidations of substituted nitrotoluenes were subsequently assessed for the relative share of CH3-group oxidation vs dioxygenation quantified as experimental branching ratios, θj, (eq 10) using the 2H- and 13C-AKIEs derived for reference reactions with toluene and nitrobenzene, respectively. n
n
∑i = 1 xih ,E·ci /∑i = 1 xil ,E·ci
h
δ E(t ) =
(h E /l E)ref
(3)
−1 (8)
n
where δhE0 and δhE are the initial isotope signature of element E and its value measured during the reaction, respectively, and c/c0 is the fraction of remaining reactant. As shown previously,16,37 data from replicate experiments were combined using the Pitman estimator.40 For the interpretation of the data modeling approach (see below), selected ϵH and ϵC values were converted to 2H- and 13C-AKIEs using an extended version of eq 2 as shown in eqs S1 and S2 of the Supporting Information. Two-dimensional H vs C isotope fractionation trends were evaluated from linear regression of logarithmic H and C isotope signatures changes as shown in eq 4 to account for large H isotope fractionation associated with CH3-group oxidation. The equivalence of eq 4 with the more widely used eq 1 for situations, in which changes of H isotope fractionation do not exceed a few hundred ‰, is shown in eqs S3−S10. The slope from this linear regression, ΛH/C, corresponds to the ratio of bulk enrichment factors, ϵH/ϵC.
ctot(t ) =
∑ ci i=1
θj =
■
(9)
kjl ∑j kjl
(10)
COMPUTATIONAL METHODS All calculations were performed with the Gaussian G09 package rev. A.0242 using the IEFPCM/B3LYP/6-31+G(d,p)43−52 level of theory based on previous studies.33,34,53 Oxidants were modeled in a singlet state. In the case of CH3-group oxidations, an unrestricted open shell method54 was applied. Optimizations were carried out with default convergence criteria and characterized by vibrational analysis. Transition states were located using the Berny algorithm.55 The intrinsic reaction 13461
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
coordinate (IRC) procedure56 was applied to investigate reaction pathways whose end points were then optimized to reactants or products. Vibrational analyses confirmed that obtained geometries (transition states, reactants, or products) corresponded to stationary points on the potential energy surfaces and also provided thermal corrections essential for evaluation of Gibbs reaction free energies. The obtained Hessians were used to calculate kinetic isotope effects (KIEs) with the ISOEFF package.57 2H-, 13C-, and 15N-kinetic isotope effects were obtained according to the Bigeleisen equation58 implemented in ISOEFF57 for each atom in each reaction of the substrate. The latter included MnO4− attack at the CH3group (denoted as Cm,a in Scheme 1) as well as on any of the six positions of the aromatic ring (denoted as Ca−Ca′), where a and a′ denote the single C atoms as in Scheme 1). H atoms are labeled according to the C atom to which they are bound to as HCa. The theoretical branching ratio for oxidation of the CH3group vs different positions of the aromatic ring, θp*, was derived from relative rate constants obtained from the Gibbs free energies of activation, ΔG⧧. Subscript p stands for the position of MnO4− attack, that is Cm,a for the CH3-group and Ca−Ca′ for positions at the aromatic ring. An asterisk (∗) is used to distinguish theoretical from experimental branching ratios (eq 10). From the computed position-specific KIEs of each element E (KIEE,p) for either CH3-group oxidation or dioxygenation, respectively, and the theoretical branching ratios, θ*p , we derived averaged atom-specific 2H-, 13C-, and 15 N-KIE values (KIEE,a, eq 11). The latter were used to derive theoretical bulk isotope enrichment factors, ϵ*E in eq 12 and compare experimentally observed isotope fractionation with theory in addition to interpretation of AKIE and KIE values. np
KIE E, a =
∑ θp*·KIEE,p 1
ϵ*E =
(11)
1 KIE E, a − 1
(12)
where np stands for the number of reactive positions of MnO4− attack. Position specific 2H-, 13C-, and 15N-KIE, averaged values (KIEE,a) and ϵE* of toluene and nitrobenzene are shown below whereas those for mono- and dinitrotoluenes are compiled in Tables S8−S12.
Figure 1. H and C isotope fractionation associated with the CH3group oxidation of toluene (upper panel) and dioxygenation of nitrobenzene (lower panel) by MnO4− (data from replicate experiments).
■
RESULTS AND DISCUSSION Isotope Fractionation Associated with CH3-Group Oxidation and Dioxygenation. Toluene CH3-Group Oxidation. Figure 1 illustrates that oxidation of toluene by MnO4− was accompanied by substantial C and H isotope fractionation. Bulk isotope enrichment factors, ϵC and ϵH (eq 3), amount to −6.0 ± 0.4 and −222 ± 9, respectively (Table 1; Figures S2 and S3), and represent large values for compounds containing seven C and eight H atoms.59 These observations agree with a large primary 2H-KIE of 9.7 reported for d8-toluene (kC7H8/ kC7D8), which was interpreted as hydride (H-) transfer mechanism from the CH3-group of toluene to MnO4− and concurrent stabilization of the carbocation by water (Scheme 1).21 We also found a second-order rate constant of 3.0 · 10−4 M−1 s−1 (Table 1) that compares well with previous data (kMnO4− between 1.0 · 10−4 and 8.3 · 10−4 M−1 s−1).31 On the basis of the ϵH value obtained from evaluation of H isotope
fractionation, however, the derivation of 2H-AKIEs with eq 2 with n = 8 H atoms would result in a meaningless negative isotope effect. Using a kinetic model to account explicitly for the reaction rates of 16 different C and H isotopomers containing not more than one heavy C or one heavy H isotope (see eqs 6 and 7 and section S3 for details), we quantified the 2H-, 13C-, and 15NAKIEs from the observed H, C, and N isotope fractionation (Table 1). The 2H- and 13C-AKIEs of 4.033 ± 0.20 and 1.0324 ± 0.0011 adequately describe the substantial isotope fractionation lending further support to the assumption that toluene oxidation occurs exclusively at the CH3-group (Figure 1). We interpret the difference of our 2H-AKIE with that measured by Gardner and Mayer21 as a consequence of secondary isotope effects of the additional aliphatic deuterium atoms in fully deuterated toluene. As shown, for example, for 13462
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
Table 1. H, C, and N Isotope Enrichment Factors (ϵH, ϵC, ϵN), ΛH/C Values, Apparent 2H-, 13C-, 15N-Kinetic Isotope Effects, CH3-Group Oxidation and Dioxygenation Branching Ratios (θm and θd), and Second-Order Rate Constants (kMnO4−) for the Oxidation of Toluene, Nitrobenzene, and Four Nitrotoluenes by MnO4− a Theoretical bulk H, C, and N isotope enrichment factors (ϵH *, ϵC*, ϵN*) as well as CH3-group oxidation and dioxygenation branching ratios (θm * and θd*). parameter
toluene
NB
2-NT
4-NT
2,4-DNT
2,6-DNT
experimental data (‰) ϵH ϵC (‰) ϵN (‰) ΛH/C (−) 2 H-AKIE (−) 13 C-AKIE (−) 15 N-AKIE (−) (10−6 M−1s−1) kMnO4−
−222 ± 9 −6.0 ± 0.4 − 34.0 ± 2.2 4.033 ± 0.20 1.0324 ± 0.0011 − 304 ± 10
14 ± 1 −9.1 ± 0.4 −1.8 ± 0.2 −1.3 ± 0.3 0.9410 ± 0.0030 1.0289 ± 0.0003 1.0017 ± 0.0003 1.6 ± 0.05
−240 ± 8 −8.8 ± 0.1 −0.8 ± 0.1 25.6 ± 1.7
−238 ± 7 −7.7 ± 0.2 −2.3 ± 0.2 28.7 ± 1.5
−76 ± 2 −8.8 ± 0.2 −2.2 ± 0.3 7.8 ± 0.5
−157 ± 6 −9.3 ± 0.3 −1.4 ± 0.1 15.3 ± 0.7
51 ± 0.4
133 ± 2
76 ± 1
18 ± 0.2
(−) (−)
1.0 0
0 1.0
0.87 ± 0.04 0.13 ± 0.04
0.95 ± 0.03 0.05 ± 0.03
0.38 ± 0.04 0.62 ± 0.03
0.58 ± 0.03 0.42 ± 0.03
(‰) (‰) (‰) (−) (−)
−285 −4.3 − 1.00 0
23 −9.5 −3.7 0 1.00
−274 −5.9 −4.1 0.93 0.07
−192 −6.2 −3.0 0.62 0.38
−89 −7.6 −3.4 0.21 0.79
−193 −7.4 −3.7 0.68 0.32
θm θd DFT calculations ϵH* b ϵ*C ϵ*N θ*m c θ*d
Uncertainties are given as ±1σ. bAll ϵ*E derived with eq 12. cθ*m corresponds to θ*p for oxidation at the Cm position shown in Table 2, and θ*d is the sum of θp* for positions C1−C2, C2−C3, and C3−C4. a
Table 2. Theoretical Evaluation of Toluene and Nitrobenzene Oxidation by MnO4− Using DFT Calculations at the IEFPCM/ B3LYP/6-31+G(d,p) Levela Gibbs free energies of reaction (ΔGR) and branching ratios for MnO4− attack at reactive position (θp*), 2H-, 13C-, and 15N-KIEs. toluene parameter ΔGR θ*p 2 H-KIE
a
(kcal mol−1) (−) (−)
13
C-KIE
(−)
15
N-KIE
(−)
nitrobenzene
p=
Cm
C1−C2
C2−C3
C3−C4
averagea
C1−C2
C2−C3
C3−C4
averagea
H C2
−20.16 0.997 1.0050
2.41 0.001 0.9032
0.94 0.002 0.9130
3.41 0 1.0128
1.0047
−8.29 0.111 0.8756
−6.50 0.790 0.9433
−2.85 0.099 0.8823
H C3
0.9959
1.0192
0.9297
0.9105
0.9087
0.9958
1.0332
1.0084
1.0310
1.0134
H C4
1.0030
H C5
0.9948
1.0028
1.0219
0.9083
1.0030
1.0185
1.0136
1.0082
1.0136
1.0073
1.0219
1.0184
0.9949
1.0146
0.8970
0.9557
H C6
0.9159
1.0039
1.0194
1.0106
1.0095
1.0039
1.0190
1.0146
1.0174
1.0154
HCm1
4.0650
1.0117
0.9839
0.9845
4.0558
HCm2
1.1186
1.0229
1.0423
1.0401
1.1184
HCm3
1.0191
0.9543
0.9939
0.9892
1.0190
average C1 C2 C3 C4 C5 C6 Cm average N C1
1.4007 1.0005 1.0010 1.0026 1.0009 1.0006 1.0012 1.0236 1.0043
0.9926 1.0243 1.0253 1.0003 1.0014 1.0012 1.0008 0.9982 1.0074
0.9873 0.999 1.0252 1.0237 1.0007 1.0013 1.0011 0.9997 1.0074
0.9839 1.0008 1.0004 1.0239 1.0246 1.0008 1.0016 0.9999 1.0074
1.3994 1.0005 1.0011 1.0026 1.0009 1.0006 1.0012 1.0235 1.0044
0.9922 1.0121 1.0362 1.0014 1.0012 1.0018 1.0012
0.9754 1.0028 1.0331 1.0171 1.0019 1.0028 1.0009
0.9789 1.0018 1.0010 1.0156 1.0342 1.0002 1.0010
0.9776 1.0037 1.0303 1.0152 1.0050 1.0024 1.0009
1.0090 1.0033
1.0098 1.0037
1.0090 1.0041
1.0096 1.0037
Average atom-specific kinetic isotope effects, KIEE,a are calculated from the weighted average of position-specific MnO4− attack as in eq 11.
predominantly at the CH3-group (θ*Cm > 99%) given the lowest ΔG⧧ for this mechanisms.53 This is also in agreement with what has been found earlier on hydride transfer mechanisms of aromatic hydrocarbon oxidation by transition metals.21,61,62 Transition states for dioxygenation of the aromatic ring are at least 3 kcal mol−1 higher in ΔG⧧ and thus negligible.53 The
the oxidation of CH4 and CH3OH, 2H-KIEs increase with the increasing replacement of deuterium for hydrogen at the reactive position.22,60 Further support for the above interpretation was obtained from theoretical examination of toluene oxidation by MnO4−. As shown in Table 2, we found that toluene is oxidized 13463
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
the dioxygenation of nitrobenzene was confined between 1.027 and 1.029 (Table 1; Table S1). DFT calculations reveal an almost identical ϵC value of −9.5‰. Notice, however, that dioxygenation is associated with two distincly different primary 13 C-KIEs ranging from 1.033 to 1.036 and 1.012 to 1.017, respectively. This phenomenon is independent of the position of MnO4− attack (Table 2) as reported earlier for benzene and phenol oxidation.33 The 13C-AKIE value of 1.0289 ± 0.0003 is slightly higher than the average of the computed 13C-KIEs for the reactive positions (C2 and C3, 1.0251), while measured and computed ϵC are identical. This difference illustrates the fact that during the derivation of AKIE values from isotope fractionation, secondary isotope effects are generally neglected leading to a slight overestimation of AKIEs at the reactive position. Quantifying Competing Oxidation Pathways of Substituted Nitrotoluenes. Reaction rate constants for the oxidation of 2- and 4-nitrotoluene as well as 2,4- and 2,6dinitrotoluene by MnO4− were smaller than those reported for toluene and larger than for nitrobenzene suggesting that the disappearance of substituted nitrotoluenes occurred by CH3group oxidation and dioxygenation simultaneously (Table 1; Figure S5). In fact, oxidations of CH3-groups in substituted toluenes by MnO4− are only moderately susceptible to substituent effects.21 Our data for toluene and 4-NT are consistent with the observation that the effect of a p-NO2-group in 4-NT did decrease the reaction rate constant by less than 1 order of magnitude. Even the presence of two NO2-groups did not reduce the rate of CH3-group oxidation below that for
magnitude of the calculated isotope effects also reflect the hydride transfer mechanism for the oxidation of the CH3-group with 2H- and 13C-KIEs of 4.0558 and 1.0235 averaged for all reactions at the reactive H and C atoms, respectively (HCm1 and Cm, Table 2). Such large 2H-KIEs would not have been observed if MnO4− would have attacked the aromatic ring. 2HKIEs for dioxygenation are all below 1.022 (see 2H-KIEs for HC2 to HC6 in Table 2). 13C-KIEs around 1.025, in contrast, could also have resulted from CH3-group oxidation and dioxygenation of the C1−C2 and C2−C3 position. The comparison between experimental AKIEs and theoretical KIEs confirms the assignment of the CH3-group oxidation mechanism as well as the consistence of experimental and computational data. Theoretical H isotope fractionation is slightly larger than the measured one as becomes evident from the bulk compound H isotope enrichment factors, ϵH (−285‰ vs −222‰, Table 1). Difference between theory and experiment vanishes when calculating the 2H-AKIE value, which is very close to the theoretical one (4.0650, Table 2), as one neglects secondary isotope effects when deriving AKIE values with eq 2 from measured isotope fractionation. This explanation also applies to the comparison of ϵC values, and 13 C-KIEs and 13C-AKIEs, respectively. Here, experimentally observed C isotope fractionation exceeds the theoretical one, and the difference becomes more pronounced when considering position-specific isotope effects (13C-AKIE of 1.0324 ± 0.0011 vs 13 C-KIEs 1.0235). As shown by the small uncertainties of AKIE and ϵE values derived with the kinetic model and linearized isotope fractionation equation (eq 3; Figure S2), this mismatch cannot be attributed solely to experimental uncertainties and remains elusive. Dioxygenation of Nitrobenzene. Fractionation of H and C isotopes during the oxidation of nitrobenzene by MnO4− is shown in Figure 1. The reaction at the aromatic ring is approximately 300 times slower than CH3-group oxidation in toluene. The bimolecular oxidation rate constant, kMnO4−, of 1.6 · 10−6 M−1 s−1 (Table 2) is almost identical with the one reported for benzene oxidation by MnO4−.31 While the extent of C isotope fractionation is very similar to that observed in toluene, H isotope fractionation is small and inverse (Figure 1). The corresponding C and H enrichment factors are −9.1 ± 0.4‰ and +14 ± 1‰, respectively (Table 2). C−H hybridization changes from sp2(C−H) to sp3(C−H) associated with dioxygenation of aromatic and olefinic moieties, as illustrated in Scheme 1, are indeed characterized by normal 13 C- and inverse secondary 2H-KIEs.29,63 The inverse effect follows from a larger frequency increase in the C−H out-ofplane bending modes relative to that of C−D.17 These effects are all reflected in the experimental AKIE and theoretical KIE values shown in Tables 1 and 2. We found that the 2H-AKIEs (0.941 ± 0.003) derived with the kinetic model returns a similar value to that derived from the ϵH of +14‰ with eqs S1 and S2 (0.965 ± 0.003, see section S2.1 and Tables S1−S3). Both experimental numbers agree well with the averaged computed atom-specific inverse 2H-KIEas for HC2 and HC5 of 0.930 and 0.916 (Table 2), which converts to a compoundaverage ϵH of +23‰. The same agreement between experiment and theory was found for the quantification of C isotope enrichment. Regardless of whether 13C-AKIE values were derived with the isotopomer-sensitive, numerical model or ϵC, its final value for
Figure 2. Linearized H vs C isotope fractionation of toluene, nitrobenzene, and four substituted nitrotoluenes during oxidation with MnO4−. Solid lines were fit with eq 4, and their slope corresponds to ΛH/C whose values are listed in Table 1. Legend entries are grouped according to decreasing ΛH/C. 13464
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
KIEs of mono- and dinitrotoluenes, which were assumed to correspond to that of nitrobenzene. Theoretical analysis of the CH3-group oxidation and dioxygenation KIEs for substituted mono- and dinitrotoluenes (Tables S8−S12) shows that these assumptions are justified. Position-specific 13C-KIEs for CH3-group oxidation of the four mono- and dinitrotoluenes only vary between 1.029 and 1.034 at the Cm-atom while 2H-KIEs span from 3.04 to 3.99 at the HCm.53 13C-KIEs and 2H-KIEs for the atoms involved in the dioxygenation averaged over the reactive positions of the aromatic ring (i.e., Ca−Ca′ and HCa) range from 1.024 to 1.028 and 0.90 to 0.94, respectively. A sensitivity analysis on the impact of KIE-variations on changes of branching ratios, θj, indicated that only a decrease of the CH3-group oxidation 2HKIEs in the above range leads to a noticeable increase in θd (and concomitant reduction of θm) by up to 0.05. Changes of 13 C-KIE for CH3-group oxidation and dioxygenation in the indicated range, in contrast, did not affect the outcome to the same extent. We therefore conclude that the combined uncertainty of the experimental branching ratios amount to 9% (0.04 + 0.05). Implications for Isotope Fractionation Analysis of Oxidative (Bio)Degradation. Many CSIA studies on micropollutant degradation such as for the aerobic and anaerobic oxidation of alkyl groups of aromatic hydrocarbons11,16,65,66 or oxidative N-dealkylation26 have reported large shifts in δ2H associated with C−H oxidations. Our results illustrate that the proposed isotopomer-sensitive modeling approach and the comparison of logarithms of δ2H vs δ13C are well suited to assess such reactions exhibiting large H isotope fractionation more accurately. Moreover, the isotopomer-sensitive model enables ones to compare isotope fractionation of compounds that react along the same reaction mechanism(s) even if they exhibit a different number of C and H atoms. (1) 2H-AKIE values derived with eq 2 (or equivalent algebraic expressions6 based on the isotope enrichment factor at the reactive position, ϵE,rp, and corrections for intramolecular isotopic competition, z) are likely to become very large once n·ϵH or z·ϵH,rp approach −1. Consequently, AKIEs might be overestimated or even inaccessible, as shown for the MnO2catalyzed oxidation of aromatic N-methyl amines.26 In addition, experimental errors scale with the factor used for isotopic dilution (n) and intramolecular competition (z) leading to very large experimental uncertainties. As shown from the results in Table 1, this problem can be circumvented with alternative procedures for AKIE estimates as the one proposed here. (2) Because δ2H vs δ13C correlations become curved instead of linear for reactions exhibiting large 2H-KIEs the more data points at large extent of substrate conversion are included, the correlation slope is shifted toward larger values and might thus no longer be indicative for a reaction pathway. The consequences can be illustrated for toluene CH3-group oxidation by MnO4− through the comparison of linear correlations between δ2H and δ13C in Figure S4 based on eq 3 with correlations of logarithmic data as ln([δ2H + 1]/[δ2H0 + 1]) vs ln([δ13C + 1]/[δ13C0 + 1]) in Figure 2 based on eq 4. The correct correlation slopes, ΛH/C, obtained from eq 4, are 34 ± 2.2 (Table 1), and, based on the ratio of ϵH/ϵC with data from Table 1 and eq 5, 37 ± 9.0. In contrast, the correlation of the nonlogarithmic data with eq 3 results in a value of 65, which substantially overestimates ϵH/ϵC. In fact, a similarly large ΛH/C value of 53 ± 5 was reported for enzymatic CH3-group
nitrobenzene, which is representing the dioxygenation pathway (Figure S5). This interpretation of mono- and dinitrotoluene oxidation kinetics is supported by the observed C and H isotope fractionation trends. While the ϵC and ϵH values in part exceed that for toluene and nitrobenzene (Table 1), the typical plot of δ13C vs δ2H (Figure S4) illustrates that the combined C and H isotope fractionation behavior of mono- and dinitrotoluenes can be interpreted as a combination of that of toluene and nitrobenzene, respectively. Data for 2-NT and 4-NT are closer to trends for toluene, implying a larger share of CH3-group oxidation than dioxygenation in contrast to 2,4-DNT and 2,6DNT. Owing to the large 2H-KIE associated with CH3-group oxidation, however, the correlated C and H isotope fractionation plot is nonlinear thus compromising the quantification of the two oxidation pathways from a linear regression of eq 1. As illustrated with the comparison of Figure 2 and Figure S4, such a comparison is no longer possible when H isotope fractionation is large and data points at large fractional extent of substrate conversion are included in the correlation or when working with isotopically enriched materials.19,64 Using a relationship based on the natural logarithms of C and H isotope fractionation as in eq 4, in contrast, enables the quantitative interpretation of linear regression slopes, ΛH/C, as ratio of isotope enrichment factors regardless of the magnitude of isotopic enrichment. The results of this analysis is shown in Figure 2 and Table 1. All ΛH/C correspond well to ratios of ϵH/ϵC. Even though the ΛH/C values derived from eq 4 enable the quantification of the relative shares of reaction pathways, they cannot be applied quantitatively to the present case due to the different number of C and H atoms in the analyzed compounds. Instead, we computed the relative shares of CH 3 -group oxidation vs dioxygenation in mono- and dinitrotoluenes from the isotopomer-specific model (eqs 6−10). As implied by the relative slopes and quantified in Table 1, 87% to 95% of mononitrotoluene transformation occurred by CH3-group oxidation (θm values of 0.87 and 0.95 for the two NTs), whereas the contribution of this pathway was only 38% and 58% in 2,4-DNT and 2,6-DNT, respectively. This result is again in good agreement with theoretical considerations, which resulted in similar shares of CH3-group oxidation and dioxygenation. Except for 4-NT, theoretical branching ratios, θj*, agreed to experimental θj values within 17%. * and θd*, were Notice that theoretical branching ratios, θm derived from the computed free energies of activation, ΔG⧧, of each pathway. Computational evidence is thus based on an isotope-independent characterization of the transition states and reaction mechanisms and allows us to confirm the validity of the experimental approach based on C and H isotope fractionation. While the experimental branching ratios are precise (i.e., standard errors of θj were
Isotopic Analysis of Oxidative Pollutant Degradation Pathways Exhibiting Large H Isotope Fractionation Reto S. Wijker,†,‡,∥ Pawel Adamczyk,§,∥ Jakov Bolotin,† Piotr Paneth,*,§ and Thomas B. Hofstetter*,†,‡ †
Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-8600 Dübendorf, Switzerland Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zürich, CH-8092 Zürich, Switzerland § Switzerland Institute of Applied Radiation Chemistry, Technical University of Lodz, Zeromskiego 116, 90-924 Lodz, Poland ‡
S Supporting Information *
ABSTRACT: Oxidation of aromatic rings and its alkyl substituents are often competing initial steps of organic pollutant transformation. The use of compound-specific isotope analysis (CSIA) to distinguish between these two pathways quantitatively, however, can be hampered by large H isotope fractionation that precludes calculation of apparent 2H-kinetic isotope effects (KIE) as well as the process identification in multi-element isotope fractionation analysis. Here, we investigated the C and H isotope fractionation associated with the transformation of toluene, nitrobenzene, and four substituted nitrotoluenes by permanganate, MnO4−, to propose a refined evaluation procedure for the quantitative distinction of CH3group oxidation and dioxygenation. On the basis of batch experiments, an isotopomer-specific kinetic model, and density functional theory (DFT) calculations, we successfully derived the large apparent 2H-KIE of 4.033 ± 0.20 for the CH3-group oxidation of toluene from H isotope fractionation exceeding >1300‰ as well as the corresponding 13C-KIE (1.0324 ± 0.0011). Experiment and theory also agreed well for the dioxygenation of nitrobenzene, which was associated with 2H- and 13C-KIEs of 0.9410 ± 0.0030 (0.9228 obtained by DFT) and 1.0289 ± 0.0003 (1.025). Consistent branching ratios for the competing CH3-group oxidation and dioxygenation of nitrotoluenes by MnO4− were obtained from the combined modeling of concentration as well as C and H isotope signature trends. Our approach offers improved estimates for the identification of contaminant microbial and abiotic oxidation pathways by CSIA.
■
and/or aromatic ring dioxygenation, respectively.10−16 The diagnostic value of such multi-element isotope fractionation patterns is based on the fact that C and H isotope fractionation is caused by kinetic isotope effects (KIEs), which are indicative of each oxidation pathway.5,6,8,17,18 The impact of 13C- and 2HKIEs on the observable isotope fractionation, however, is diluted by nonreactive atoms in a molecule. Dilution of isotope fractionation is taken into account in bulk compound C and H isotope enrichment factors (ϵC and ϵH, respectively), which are used to quantify a contaminant’s isotope fractionation behavior.6 Correlations of changing H and C isotope signatures, δ2H vs δ13C, can be used for the identification and quantification of competing reaction pathways because the correlation slopes, Δδ2H vs Δδ13C (eq 1), follow approximately the ratio of ϵC and ϵH values.19,20
INTRODUCTION Oxidations of methyl groups and aromatic rings are frequent initial steps in the enzymatic and abiotic degradation of persistent aromatic contaminants in the environment such as nitroaromatic explosives and fuel components (BTEX).1,2 Because both processes can occur concurrently and lead to products that are susceptible for further transformation, they are difficult to track, thus making a quantitative assessment of such transformations very challenging.3 This problem was successfully alleviated by the application of compound-specific isotope analysis (CSIA), with which changes of isotope ratios in the residual contaminant have been used for quantifying the extent of transformation even if multiple chemical and/or biological reactions took place.4−9 Because oxidation of methyl groups and aromatic rings involve the cleavage of C−H bonds or CC and C−H hybridization changes, the combined analysis of changes in C and H isotope ratios is a robust measure to assess (bio)degradation without explicit need for monitoring reactant and product concentration time series. Indeed, laboratory and field studies with various substituted benzenes have revealed characteristic C and H isotope fractionation trends for contaminant degradation initiated via methyl group oxidation © 2013 American Chemical Society
Received: Revised: Accepted: Published: 13459
August 12, 2013 October 25, 2013 October 31, 2013 October 31, 2013 dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
Scheme 1. Initial Steps of MnO4− Catalyzed Oxidation Pathways of Toluene and Nitrotoluenes, Including Their C and H Atom Labels Used Throughout the Articlea
a
Water is involved in the stabilization of the CH3-group oxidation transition state.21
ϵ Δδ 2 H ≈ H ϵC Δδ13C
fractionation patterns of aliphatic and aromatic C oxygenation of six model contaminants by permanganate (MnO4−). Based on the comparison of (experimental) 2H- and 13C-AKIE values with the corresponding (computed) intrinsic KIEs, we propose a new data evaluation procedure for (i) the derivation of large 2 H-AKIEs from H isotope fractionation data as well as (ii) the interpretation of competing CH3-group oxidation and dioxygenation reactions in typical two-dimensional C vs H isotope fractionation plots for compounds that differ in the number of C and H atoms. MnO4− was chosen as model oxidant owing to its use for in situ chemical oxidation of organic contaminants31 as well as previous experimental and theoretical reports on large KIEs associated with aliphatic and olefinic C−H bond oxygenations19,21,29,32−34 (Scheme 1). The set of probe compounds included typical soil and water contaminants that react with MnO4− exclusively via CH3-group oxidation (toluene21) and aromatic ring dioxygenation (nitrobenzene), respectively, as well as four nitro- and dinitrotoluenes (2- and 4-nitrotoluene, 2,4- and 2,6-dinitrotoluene) for which both oxygenation pathways occur.
(1)
2
However, H-KIEs associated with the cleavage of aliphatic C−H bonds can be substantial (i.e., 7−5019,21−25) and eq 1 no longer holds because δ2H vs δ13C correlations become nonlinear as pointed out earlier by Elsner et al.19 This phenomenon compromises the identification of reaction pathways by CSIA in dual isotope plots. Moreover, the large H isotope fractionation arising from such reactions cannot be quantified as apparent kinetic isotope effects (AKIE). The typically used AKIE-approximation, eq 2, leads to numerically erroneous results including meaningless negative AKIE values once bulk H isotope enrichment factors, ϵH, approach the negative inverse of the number of H atoms, −1/n, in a compound.26 AKIE E ≈
1 1 + n·ϵE
(2)
where subscript E stands for isotopic elements H and C, respectively. Failure to compare intrinsic 2H-KIEs with experimental 2H-AKIEs derived from H isotope fractionation as in eq 2 impedes a mechanistic interpretation of the observed contaminant isotope fractionation27 that would be necessary for a correct application of CSIA. Another challenge regarding the interpretation of AKIEs obtained with eq 2 arises for dioxygenation pathways. Such concerted reactions exhibit two, often distinctly different primary 2 H- and 13C-KIEs due to asynchronous dioxygen addition.15,28,29 Because eq 2 returns the weighted average of two position-specific KIEs,30 a comparison of 13C-AKIEs for dioxygenation warrants theoretical support such as density functional theory-based KIE calculations for unambiguous interpretation. The goal of this study was to contribute to a precise interpretation of C and H isotope fractionation data gained from CSIA for oxidation reactions of substituted aromatic contaminants. To this end, we performed a combined experimental and computational study in which we evaluated the 2H- and 13C-kinetic isotope effects and C and H isotope
■
EXPERIMENTAL METHODS A complete list of all chemicals used and their purity can be found in the Supporting Information (SI). Batch Experiments. Oxidation experiments with toluene (1.8 mM initial concentration), 2-NT (1.9 mM), 4-NT (1.3 mM), and 2,4-DNT (0.5 mM) were carried out in batch reactors containing 50 mM KMnO4. Because of slow turnover, oxidations of nitrobenzene (1.9 mM) and 2,6-DNT (0.4 mM) were performed in 300 mM and 87.5 mM KMnO4, respectively. All reactions took place at room temperature in serum bottles closed with Viton stoppers and buffered at pH 7.2 with 25 mM potassium phosphate. Blanks were treated identically except for the addition of KMnO4 to rule out losses due to phase transfer processes. Oxidations were initiated through the addition of KMnO4 stock solution. At predefined time points, aqueous samples were withdrawn with a gastight glass syringe, and the reaction was stopped by the addition of stoichiometric excess of NaS2O3. The samples were subsequently filtered through a 0.22 13460
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
⎛ δ13C + 1 ⎞ ⎛ δ 2H + 1 ⎞ ϵ ⎟ ⎟ = H ·ln ⎜ 13 ln ⎜ 2 ϵC ⎝ δ C0 + 1 ⎠ ⎝ δ H0 + 1 ⎠
μm hydrophilic PTFE filter and stored at 4 °C until chemical and isotopic analysis. Chemical and Isotopic Analyses. Concentrations of nitrobenzene as well as mono- and dinitrotoluenes were quantified by reverse phase HPLC (Dionex UltiMade 3000) using a Supelcosil LC-18 column (25 cm × 4.6 mm, 5 μm, Supelco) and UV−vis detection at wavelengths corresponding to the absorption maxima of the compounds. The eluent mixture consisted of 65% methanol and 35% H2O at a flow rate of 1 mL min−1. Toluene was extracted from buffered aqueous solution in hexane and measured by GC/MS using methods described recently.10 Stable C, H, and N isotope signatures of toluene and nitrotoluenes were measured using a gas chromatograph coupled to an isotope ratio mass spectrometer via a combustion interface as reported previously.35−37 Briefly, extraction and enrichment from aqueous samples containing nitrotoluenes or nitrobenzene were performed by solid-phase microextraction (SPME, 85 μm polyacrylate coating, Supelco) at 40 °C for 45 min. Toluene was extracted from a purge and trap system (Teledyne Tekmar) according to the method reported in ref 38. Samples were diluted to concentrations yielding similar peak amplitudes (between 1 and 5 V) to minimize analytical uncertainties due to instrument nonlinearity. C, N, and H isotope signatures (δ13C, δ15N, and δ2H), are reported as arithmetic mean of triplicate measurements (±σ) in per mil (‰) relative to Vienna PeeDee Belemnite (δ13CVPDB), air (δ15Nair), and Vienna standard mean ocean water (δ2HVSMOW), respectively. All compounds were available as calibrated inhouse standards, whose δ 13 C and δ 15 N values were characterized independently by elemental analyzer IRMS and used in standard bracketing procedures to ensure accuracy of isotopic measurements. Hydrogen isotope ratios were calibrated using different alkanes and dimethylaniline as reference compounds39 in the δ2H-range between −50‰ to 500‰. Evaluation of Experimental Data. For the comparison of data evaluation procedures, typical bulk isotope enrichment factors, ϵE, were derived from linear regression of δ2H or δ13C values versus the remaining fraction of the reactant as shown in eq 3. ⎛ δhE + 1 ⎞ ⎛c⎞ ⎟⎟ = ϵE ·ln ⎜ ⎟ ln ⎜⎜ h ⎝ c0 ⎠ ⎝ δ E0 + 1 ⎠
ΛH/C =
ϵH ϵC
(4)
(5)
2 H- and 13C-AKIE values pertinent to CH3-group oxidation and dioxygenation were calculated from experiments with toluene and nitrobenzene as substrates, respectively. To this end, we implemented a numerical model in Aquasim41 consisting of a set of differential equations for all C and H isotopologues and isotopomers containing light and one heavy C or H isotope, respectively (eq 6).
d[ci] = −∑ γjE·kjE·[ci] dt j
AKIEEj
(6)
kjl ,E
=
kjh ,E
(7)
where [ci] is the concentration of isotopomer i (see compilation of isotopomers in section S3 of the Supporting Information), γEj is the probability of reaction j (i.e., CH3-group oxidation or dioxygenation) at a reactive position of the substrate, and kEj is the reaction’s second-order rate constant. γEj is used to account explicitly for intramolecular competition of randomly distributed heavy isotopes (Tables S4−S7, Supporting Information). Superscript E stands for the C and H isotopes, respectively. 2H- and 13C-AKIE values follow from the ratio of reaction rate constants of light (l) and heavy (h) C and H isotopomers (eq 7). The numerical model was fit to the evolution of measured isotope signatures, δhE(t), and substrate concentrations, ctot(t), over the time course of the reactions. These two variables were obtained numerically from summing up n isotopomers as well as C and H atoms (xh,E and xl,E i i ) as shown in eqs 8 and 9. − MnO4 -catalyzed oxidations of substituted nitrotoluenes were subsequently assessed for the relative share of CH3-group oxidation vs dioxygenation quantified as experimental branching ratios, θj, (eq 10) using the 2H- and 13C-AKIEs derived for reference reactions with toluene and nitrobenzene, respectively. n
n
∑i = 1 xih ,E·ci /∑i = 1 xil ,E·ci
h
δ E(t ) =
(h E /l E)ref
(3)
−1 (8)
n
where δhE0 and δhE are the initial isotope signature of element E and its value measured during the reaction, respectively, and c/c0 is the fraction of remaining reactant. As shown previously,16,37 data from replicate experiments were combined using the Pitman estimator.40 For the interpretation of the data modeling approach (see below), selected ϵH and ϵC values were converted to 2H- and 13C-AKIEs using an extended version of eq 2 as shown in eqs S1 and S2 of the Supporting Information. Two-dimensional H vs C isotope fractionation trends were evaluated from linear regression of logarithmic H and C isotope signatures changes as shown in eq 4 to account for large H isotope fractionation associated with CH3-group oxidation. The equivalence of eq 4 with the more widely used eq 1 for situations, in which changes of H isotope fractionation do not exceed a few hundred ‰, is shown in eqs S3−S10. The slope from this linear regression, ΛH/C, corresponds to the ratio of bulk enrichment factors, ϵH/ϵC.
ctot(t ) =
∑ ci i=1
θj =
■
(9)
kjl ∑j kjl
(10)
COMPUTATIONAL METHODS All calculations were performed with the Gaussian G09 package rev. A.0242 using the IEFPCM/B3LYP/6-31+G(d,p)43−52 level of theory based on previous studies.33,34,53 Oxidants were modeled in a singlet state. In the case of CH3-group oxidations, an unrestricted open shell method54 was applied. Optimizations were carried out with default convergence criteria and characterized by vibrational analysis. Transition states were located using the Berny algorithm.55 The intrinsic reaction 13461
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
coordinate (IRC) procedure56 was applied to investigate reaction pathways whose end points were then optimized to reactants or products. Vibrational analyses confirmed that obtained geometries (transition states, reactants, or products) corresponded to stationary points on the potential energy surfaces and also provided thermal corrections essential for evaluation of Gibbs reaction free energies. The obtained Hessians were used to calculate kinetic isotope effects (KIEs) with the ISOEFF package.57 2H-, 13C-, and 15N-kinetic isotope effects were obtained according to the Bigeleisen equation58 implemented in ISOEFF57 for each atom in each reaction of the substrate. The latter included MnO4− attack at the CH3group (denoted as Cm,a in Scheme 1) as well as on any of the six positions of the aromatic ring (denoted as Ca−Ca′), where a and a′ denote the single C atoms as in Scheme 1). H atoms are labeled according to the C atom to which they are bound to as HCa. The theoretical branching ratio for oxidation of the CH3group vs different positions of the aromatic ring, θp*, was derived from relative rate constants obtained from the Gibbs free energies of activation, ΔG⧧. Subscript p stands for the position of MnO4− attack, that is Cm,a for the CH3-group and Ca−Ca′ for positions at the aromatic ring. An asterisk (∗) is used to distinguish theoretical from experimental branching ratios (eq 10). From the computed position-specific KIEs of each element E (KIEE,p) for either CH3-group oxidation or dioxygenation, respectively, and the theoretical branching ratios, θ*p , we derived averaged atom-specific 2H-, 13C-, and 15 N-KIE values (KIEE,a, eq 11). The latter were used to derive theoretical bulk isotope enrichment factors, ϵ*E in eq 12 and compare experimentally observed isotope fractionation with theory in addition to interpretation of AKIE and KIE values. np
KIE E, a =
∑ θp*·KIEE,p 1
ϵ*E =
(11)
1 KIE E, a − 1
(12)
where np stands for the number of reactive positions of MnO4− attack. Position specific 2H-, 13C-, and 15N-KIE, averaged values (KIEE,a) and ϵE* of toluene and nitrobenzene are shown below whereas those for mono- and dinitrotoluenes are compiled in Tables S8−S12.
Figure 1. H and C isotope fractionation associated with the CH3group oxidation of toluene (upper panel) and dioxygenation of nitrobenzene (lower panel) by MnO4− (data from replicate experiments).
■
RESULTS AND DISCUSSION Isotope Fractionation Associated with CH3-Group Oxidation and Dioxygenation. Toluene CH3-Group Oxidation. Figure 1 illustrates that oxidation of toluene by MnO4− was accompanied by substantial C and H isotope fractionation. Bulk isotope enrichment factors, ϵC and ϵH (eq 3), amount to −6.0 ± 0.4 and −222 ± 9, respectively (Table 1; Figures S2 and S3), and represent large values for compounds containing seven C and eight H atoms.59 These observations agree with a large primary 2H-KIE of 9.7 reported for d8-toluene (kC7H8/ kC7D8), which was interpreted as hydride (H-) transfer mechanism from the CH3-group of toluene to MnO4− and concurrent stabilization of the carbocation by water (Scheme 1).21 We also found a second-order rate constant of 3.0 · 10−4 M−1 s−1 (Table 1) that compares well with previous data (kMnO4− between 1.0 · 10−4 and 8.3 · 10−4 M−1 s−1).31 On the basis of the ϵH value obtained from evaluation of H isotope
fractionation, however, the derivation of 2H-AKIEs with eq 2 with n = 8 H atoms would result in a meaningless negative isotope effect. Using a kinetic model to account explicitly for the reaction rates of 16 different C and H isotopomers containing not more than one heavy C or one heavy H isotope (see eqs 6 and 7 and section S3 for details), we quantified the 2H-, 13C-, and 15NAKIEs from the observed H, C, and N isotope fractionation (Table 1). The 2H- and 13C-AKIEs of 4.033 ± 0.20 and 1.0324 ± 0.0011 adequately describe the substantial isotope fractionation lending further support to the assumption that toluene oxidation occurs exclusively at the CH3-group (Figure 1). We interpret the difference of our 2H-AKIE with that measured by Gardner and Mayer21 as a consequence of secondary isotope effects of the additional aliphatic deuterium atoms in fully deuterated toluene. As shown, for example, for 13462
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
Table 1. H, C, and N Isotope Enrichment Factors (ϵH, ϵC, ϵN), ΛH/C Values, Apparent 2H-, 13C-, 15N-Kinetic Isotope Effects, CH3-Group Oxidation and Dioxygenation Branching Ratios (θm and θd), and Second-Order Rate Constants (kMnO4−) for the Oxidation of Toluene, Nitrobenzene, and Four Nitrotoluenes by MnO4− a Theoretical bulk H, C, and N isotope enrichment factors (ϵH *, ϵC*, ϵN*) as well as CH3-group oxidation and dioxygenation branching ratios (θm * and θd*). parameter
toluene
NB
2-NT
4-NT
2,4-DNT
2,6-DNT
experimental data (‰) ϵH ϵC (‰) ϵN (‰) ΛH/C (−) 2 H-AKIE (−) 13 C-AKIE (−) 15 N-AKIE (−) (10−6 M−1s−1) kMnO4−
−222 ± 9 −6.0 ± 0.4 − 34.0 ± 2.2 4.033 ± 0.20 1.0324 ± 0.0011 − 304 ± 10
14 ± 1 −9.1 ± 0.4 −1.8 ± 0.2 −1.3 ± 0.3 0.9410 ± 0.0030 1.0289 ± 0.0003 1.0017 ± 0.0003 1.6 ± 0.05
−240 ± 8 −8.8 ± 0.1 −0.8 ± 0.1 25.6 ± 1.7
−238 ± 7 −7.7 ± 0.2 −2.3 ± 0.2 28.7 ± 1.5
−76 ± 2 −8.8 ± 0.2 −2.2 ± 0.3 7.8 ± 0.5
−157 ± 6 −9.3 ± 0.3 −1.4 ± 0.1 15.3 ± 0.7
51 ± 0.4
133 ± 2
76 ± 1
18 ± 0.2
(−) (−)
1.0 0
0 1.0
0.87 ± 0.04 0.13 ± 0.04
0.95 ± 0.03 0.05 ± 0.03
0.38 ± 0.04 0.62 ± 0.03
0.58 ± 0.03 0.42 ± 0.03
(‰) (‰) (‰) (−) (−)
−285 −4.3 − 1.00 0
23 −9.5 −3.7 0 1.00
−274 −5.9 −4.1 0.93 0.07
−192 −6.2 −3.0 0.62 0.38
−89 −7.6 −3.4 0.21 0.79
−193 −7.4 −3.7 0.68 0.32
θm θd DFT calculations ϵH* b ϵ*C ϵ*N θ*m c θ*d
Uncertainties are given as ±1σ. bAll ϵ*E derived with eq 12. cθ*m corresponds to θ*p for oxidation at the Cm position shown in Table 2, and θ*d is the sum of θp* for positions C1−C2, C2−C3, and C3−C4. a
Table 2. Theoretical Evaluation of Toluene and Nitrobenzene Oxidation by MnO4− Using DFT Calculations at the IEFPCM/ B3LYP/6-31+G(d,p) Levela Gibbs free energies of reaction (ΔGR) and branching ratios for MnO4− attack at reactive position (θp*), 2H-, 13C-, and 15N-KIEs. toluene parameter ΔGR θ*p 2 H-KIE
a
(kcal mol−1) (−) (−)
13
C-KIE
(−)
15
N-KIE
(−)
nitrobenzene
p=
Cm
C1−C2
C2−C3
C3−C4
averagea
C1−C2
C2−C3
C3−C4
averagea
H C2
−20.16 0.997 1.0050
2.41 0.001 0.9032
0.94 0.002 0.9130
3.41 0 1.0128
1.0047
−8.29 0.111 0.8756
−6.50 0.790 0.9433
−2.85 0.099 0.8823
H C3
0.9959
1.0192
0.9297
0.9105
0.9087
0.9958
1.0332
1.0084
1.0310
1.0134
H C4
1.0030
H C5
0.9948
1.0028
1.0219
0.9083
1.0030
1.0185
1.0136
1.0082
1.0136
1.0073
1.0219
1.0184
0.9949
1.0146
0.8970
0.9557
H C6
0.9159
1.0039
1.0194
1.0106
1.0095
1.0039
1.0190
1.0146
1.0174
1.0154
HCm1
4.0650
1.0117
0.9839
0.9845
4.0558
HCm2
1.1186
1.0229
1.0423
1.0401
1.1184
HCm3
1.0191
0.9543
0.9939
0.9892
1.0190
average C1 C2 C3 C4 C5 C6 Cm average N C1
1.4007 1.0005 1.0010 1.0026 1.0009 1.0006 1.0012 1.0236 1.0043
0.9926 1.0243 1.0253 1.0003 1.0014 1.0012 1.0008 0.9982 1.0074
0.9873 0.999 1.0252 1.0237 1.0007 1.0013 1.0011 0.9997 1.0074
0.9839 1.0008 1.0004 1.0239 1.0246 1.0008 1.0016 0.9999 1.0074
1.3994 1.0005 1.0011 1.0026 1.0009 1.0006 1.0012 1.0235 1.0044
0.9922 1.0121 1.0362 1.0014 1.0012 1.0018 1.0012
0.9754 1.0028 1.0331 1.0171 1.0019 1.0028 1.0009
0.9789 1.0018 1.0010 1.0156 1.0342 1.0002 1.0010
0.9776 1.0037 1.0303 1.0152 1.0050 1.0024 1.0009
1.0090 1.0033
1.0098 1.0037
1.0090 1.0041
1.0096 1.0037
Average atom-specific kinetic isotope effects, KIEE,a are calculated from the weighted average of position-specific MnO4− attack as in eq 11.
predominantly at the CH3-group (θ*Cm > 99%) given the lowest ΔG⧧ for this mechanisms.53 This is also in agreement with what has been found earlier on hydride transfer mechanisms of aromatic hydrocarbon oxidation by transition metals.21,61,62 Transition states for dioxygenation of the aromatic ring are at least 3 kcal mol−1 higher in ΔG⧧ and thus negligible.53 The
the oxidation of CH4 and CH3OH, 2H-KIEs increase with the increasing replacement of deuterium for hydrogen at the reactive position.22,60 Further support for the above interpretation was obtained from theoretical examination of toluene oxidation by MnO4−. As shown in Table 2, we found that toluene is oxidized 13463
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
the dioxygenation of nitrobenzene was confined between 1.027 and 1.029 (Table 1; Table S1). DFT calculations reveal an almost identical ϵC value of −9.5‰. Notice, however, that dioxygenation is associated with two distincly different primary 13 C-KIEs ranging from 1.033 to 1.036 and 1.012 to 1.017, respectively. This phenomenon is independent of the position of MnO4− attack (Table 2) as reported earlier for benzene and phenol oxidation.33 The 13C-AKIE value of 1.0289 ± 0.0003 is slightly higher than the average of the computed 13C-KIEs for the reactive positions (C2 and C3, 1.0251), while measured and computed ϵC are identical. This difference illustrates the fact that during the derivation of AKIE values from isotope fractionation, secondary isotope effects are generally neglected leading to a slight overestimation of AKIEs at the reactive position. Quantifying Competing Oxidation Pathways of Substituted Nitrotoluenes. Reaction rate constants for the oxidation of 2- and 4-nitrotoluene as well as 2,4- and 2,6dinitrotoluene by MnO4− were smaller than those reported for toluene and larger than for nitrobenzene suggesting that the disappearance of substituted nitrotoluenes occurred by CH3group oxidation and dioxygenation simultaneously (Table 1; Figure S5). In fact, oxidations of CH3-groups in substituted toluenes by MnO4− are only moderately susceptible to substituent effects.21 Our data for toluene and 4-NT are consistent with the observation that the effect of a p-NO2-group in 4-NT did decrease the reaction rate constant by less than 1 order of magnitude. Even the presence of two NO2-groups did not reduce the rate of CH3-group oxidation below that for
magnitude of the calculated isotope effects also reflect the hydride transfer mechanism for the oxidation of the CH3-group with 2H- and 13C-KIEs of 4.0558 and 1.0235 averaged for all reactions at the reactive H and C atoms, respectively (HCm1 and Cm, Table 2). Such large 2H-KIEs would not have been observed if MnO4− would have attacked the aromatic ring. 2HKIEs for dioxygenation are all below 1.022 (see 2H-KIEs for HC2 to HC6 in Table 2). 13C-KIEs around 1.025, in contrast, could also have resulted from CH3-group oxidation and dioxygenation of the C1−C2 and C2−C3 position. The comparison between experimental AKIEs and theoretical KIEs confirms the assignment of the CH3-group oxidation mechanism as well as the consistence of experimental and computational data. Theoretical H isotope fractionation is slightly larger than the measured one as becomes evident from the bulk compound H isotope enrichment factors, ϵH (−285‰ vs −222‰, Table 1). Difference between theory and experiment vanishes when calculating the 2H-AKIE value, which is very close to the theoretical one (4.0650, Table 2), as one neglects secondary isotope effects when deriving AKIE values with eq 2 from measured isotope fractionation. This explanation also applies to the comparison of ϵC values, and 13 C-KIEs and 13C-AKIEs, respectively. Here, experimentally observed C isotope fractionation exceeds the theoretical one, and the difference becomes more pronounced when considering position-specific isotope effects (13C-AKIE of 1.0324 ± 0.0011 vs 13 C-KIEs 1.0235). As shown by the small uncertainties of AKIE and ϵE values derived with the kinetic model and linearized isotope fractionation equation (eq 3; Figure S2), this mismatch cannot be attributed solely to experimental uncertainties and remains elusive. Dioxygenation of Nitrobenzene. Fractionation of H and C isotopes during the oxidation of nitrobenzene by MnO4− is shown in Figure 1. The reaction at the aromatic ring is approximately 300 times slower than CH3-group oxidation in toluene. The bimolecular oxidation rate constant, kMnO4−, of 1.6 · 10−6 M−1 s−1 (Table 2) is almost identical with the one reported for benzene oxidation by MnO4−.31 While the extent of C isotope fractionation is very similar to that observed in toluene, H isotope fractionation is small and inverse (Figure 1). The corresponding C and H enrichment factors are −9.1 ± 0.4‰ and +14 ± 1‰, respectively (Table 2). C−H hybridization changes from sp2(C−H) to sp3(C−H) associated with dioxygenation of aromatic and olefinic moieties, as illustrated in Scheme 1, are indeed characterized by normal 13 C- and inverse secondary 2H-KIEs.29,63 The inverse effect follows from a larger frequency increase in the C−H out-ofplane bending modes relative to that of C−D.17 These effects are all reflected in the experimental AKIE and theoretical KIE values shown in Tables 1 and 2. We found that the 2H-AKIEs (0.941 ± 0.003) derived with the kinetic model returns a similar value to that derived from the ϵH of +14‰ with eqs S1 and S2 (0.965 ± 0.003, see section S2.1 and Tables S1−S3). Both experimental numbers agree well with the averaged computed atom-specific inverse 2H-KIEas for HC2 and HC5 of 0.930 and 0.916 (Table 2), which converts to a compoundaverage ϵH of +23‰. The same agreement between experiment and theory was found for the quantification of C isotope enrichment. Regardless of whether 13C-AKIE values were derived with the isotopomer-sensitive, numerical model or ϵC, its final value for
Figure 2. Linearized H vs C isotope fractionation of toluene, nitrobenzene, and four substituted nitrotoluenes during oxidation with MnO4−. Solid lines were fit with eq 4, and their slope corresponds to ΛH/C whose values are listed in Table 1. Legend entries are grouped according to decreasing ΛH/C. 13464
dx.doi.org/10.1021/es403597v | Environ. Sci. Technol. 2013, 47, 13459−13468
Environmental Science & Technology
Article
KIEs of mono- and dinitrotoluenes, which were assumed to correspond to that of nitrobenzene. Theoretical analysis of the CH3-group oxidation and dioxygenation KIEs for substituted mono- and dinitrotoluenes (Tables S8−S12) shows that these assumptions are justified. Position-specific 13C-KIEs for CH3-group oxidation of the four mono- and dinitrotoluenes only vary between 1.029 and 1.034 at the Cm-atom while 2H-KIEs span from 3.04 to 3.99 at the HCm.53 13C-KIEs and 2H-KIEs for the atoms involved in the dioxygenation averaged over the reactive positions of the aromatic ring (i.e., Ca−Ca′ and HCa) range from 1.024 to 1.028 and 0.90 to 0.94, respectively. A sensitivity analysis on the impact of KIE-variations on changes of branching ratios, θj, indicated that only a decrease of the CH3-group oxidation 2HKIEs in the above range leads to a noticeable increase in θd (and concomitant reduction of θm) by up to 0.05. Changes of 13 C-KIE for CH3-group oxidation and dioxygenation in the indicated range, in contrast, did not affect the outcome to the same extent. We therefore conclude that the combined uncertainty of the experimental branching ratios amount to 9% (0.04 + 0.05). Implications for Isotope Fractionation Analysis of Oxidative (Bio)Degradation. Many CSIA studies on micropollutant degradation such as for the aerobic and anaerobic oxidation of alkyl groups of aromatic hydrocarbons11,16,65,66 or oxidative N-dealkylation26 have reported large shifts in δ2H associated with C−H oxidations. Our results illustrate that the proposed isotopomer-sensitive modeling approach and the comparison of logarithms of δ2H vs δ13C are well suited to assess such reactions exhibiting large H isotope fractionation more accurately. Moreover, the isotopomer-sensitive model enables ones to compare isotope fractionation of compounds that react along the same reaction mechanism(s) even if they exhibit a different number of C and H atoms. (1) 2H-AKIE values derived with eq 2 (or equivalent algebraic expressions6 based on the isotope enrichment factor at the reactive position, ϵE,rp, and corrections for intramolecular isotopic competition, z) are likely to become very large once n·ϵH or z·ϵH,rp approach −1. Consequently, AKIEs might be overestimated or even inaccessible, as shown for the MnO2catalyzed oxidation of aromatic N-methyl amines.26 In addition, experimental errors scale with the factor used for isotopic dilution (n) and intramolecular competition (z) leading to very large experimental uncertainties. As shown from the results in Table 1, this problem can be circumvented with alternative procedures for AKIE estimates as the one proposed here. (2) Because δ2H vs δ13C correlations become curved instead of linear for reactions exhibiting large 2H-KIEs the more data points at large extent of substrate conversion are included, the correlation slope is shifted toward larger values and might thus no longer be indicative for a reaction pathway. The consequences can be illustrated for toluene CH3-group oxidation by MnO4− through the comparison of linear correlations between δ2H and δ13C in Figure S4 based on eq 3 with correlations of logarithmic data as ln([δ2H + 1]/[δ2H0 + 1]) vs ln([δ13C + 1]/[δ13C0 + 1]) in Figure 2 based on eq 4. The correct correlation slopes, ΛH/C, obtained from eq 4, are 34 ± 2.2 (Table 1), and, based on the ratio of ϵH/ϵC with data from Table 1 and eq 5, 37 ± 9.0. In contrast, the correlation of the nonlogarithmic data with eq 3 results in a value of 65, which substantially overestimates ϵH/ϵC. In fact, a similarly large ΛH/C value of 53 ± 5 was reported for enzymatic CH3-group
nitrobenzene, which is representing the dioxygenation pathway (Figure S5). This interpretation of mono- and dinitrotoluene oxidation kinetics is supported by the observed C and H isotope fractionation trends. While the ϵC and ϵH values in part exceed that for toluene and nitrobenzene (Table 1), the typical plot of δ13C vs δ2H (Figure S4) illustrates that the combined C and H isotope fractionation behavior of mono- and dinitrotoluenes can be interpreted as a combination of that of toluene and nitrobenzene, respectively. Data for 2-NT and 4-NT are closer to trends for toluene, implying a larger share of CH3-group oxidation than dioxygenation in contrast to 2,4-DNT and 2,6DNT. Owing to the large 2H-KIE associated with CH3-group oxidation, however, the correlated C and H isotope fractionation plot is nonlinear thus compromising the quantification of the two oxidation pathways from a linear regression of eq 1. As illustrated with the comparison of Figure 2 and Figure S4, such a comparison is no longer possible when H isotope fractionation is large and data points at large fractional extent of substrate conversion are included in the correlation or when working with isotopically enriched materials.19,64 Using a relationship based on the natural logarithms of C and H isotope fractionation as in eq 4, in contrast, enables the quantitative interpretation of linear regression slopes, ΛH/C, as ratio of isotope enrichment factors regardless of the magnitude of isotopic enrichment. The results of this analysis is shown in Figure 2 and Table 1. All ΛH/C correspond well to ratios of ϵH/ϵC. Even though the ΛH/C values derived from eq 4 enable the quantification of the relative shares of reaction pathways, they cannot be applied quantitatively to the present case due to the different number of C and H atoms in the analyzed compounds. Instead, we computed the relative shares of CH 3 -group oxidation vs dioxygenation in mono- and dinitrotoluenes from the isotopomer-specific model (eqs 6−10). As implied by the relative slopes and quantified in Table 1, 87% to 95% of mononitrotoluene transformation occurred by CH3-group oxidation (θm values of 0.87 and 0.95 for the two NTs), whereas the contribution of this pathway was only 38% and 58% in 2,4-DNT and 2,6-DNT, respectively. This result is again in good agreement with theoretical considerations, which resulted in similar shares of CH3-group oxidation and dioxygenation. Except for 4-NT, theoretical branching ratios, θj*, agreed to experimental θj values within 17%. * and θd*, were Notice that theoretical branching ratios, θm derived from the computed free energies of activation, ΔG⧧, of each pathway. Computational evidence is thus based on an isotope-independent characterization of the transition states and reaction mechanisms and allows us to confirm the validity of the experimental approach based on C and H isotope fractionation. While the experimental branching ratios are precise (i.e., standard errors of θj were
